Download High Charge Short Electron Bunches for Wakefield Accelerator

HIGH CHARGE SHORT ELECTRON BUNCHES FOR WAKEFIELD
ACCELERATOR STRUCTURES DEVELOPMENT*
M.E. Conde, W. Gai, R. Konecny, J.G. Power, P. Schoessow
Argonne National Laboratory, Argonne, Illinois 60439 USA
generating low charge short electron bunches
synchronized with the drive bunches (witness gun), and
(iii) an experimental section for wakefield measurements,
in which the witness beam is accelerated by wakefields
generated by the drive beam. We have obtained electron
bunches of 10 - 100 nC with FWHM of 10 - 40 ps, which
have been used for initial wakefield experiments in both
dielectric loaded structures [1] and plasmas [2]. These
unprecedented high-charge short-electron bunches, allied
with the uniqueness of having two photocathode RF guns
with adjustable delay between their beams, make AWA
an ideal facility for the study of electron beam driven
wakefield acceleration.
Abstract
The Argonne Wakefield Accelerator group develops
accelerating structures based on dielectric loaded
waveguides. We use high charge short electron bunches
to excite wakefields in dielectric loaded structures, and a
second (low charge) beam to probe the wakefields left
behind by the drive beam. We report measurements of
beam parameters and also initial results of the dielectric
loaded accelerating structures. We have studied
acceleration of the probe beam in these structures and we
have also made measurements on the RF pulses that are
generated by the drive beam. Single drive bunches, as
well as multiple bunches separated by an integer number
of RF periods have been used to generate the accelerating
wakefields.
2 AWA ELECTRON GUNS AND LINAC
STRUCTURES
Figure 1 shows a schematic of the AWA layout. The
half-cell drive gun was designed to have a high
accelerating field (92 MV/m at the photocathode surface)
to allow the extraction of high charge electron bunches,
and to produce a 2 MeV beam with the limited RF power
available at the design time (1.5 MW at 1.3 GHz) [3]. The
high intensity of the accelerating field in the gun permits
the generation and acceleration of short electron bunches,
without having to rely on magnetic pulse compression.
Magnesium was chosen to be the photocathode material
-4
for its ruggedness and quantum efficiency (5×10 ). The 2
MeV bunches generated by the drive gun subsequently
1 INTRODUCTION
In order to study and demonstrate the wakefield
acceleration scheme and also the two beam acceleration
concept, we have designed and constructed a facility
called Argonne Wakefield Accelerator (AWA). The
AWA consists of three major components: (i) a
photocathode based RF electron gun capable of
producing up to 100 nC (drive gun) followed by two
standing wave linac sections for post acceleration, (ii) a
second photocathode based RF electron gun capable of
S
S
S
I4
Q
Q
Q
C6
D
C7
Witness Gun
C8
Q
Q
Q
S
S
S
C1
C2
Q
Q
Q
Test Section
D
I1
Drive Gun
D
D
Q
C3
I2
Q
Q C5
I3
Q
C4
Spectrometer
Linac Tanks
Figure 1: Schematic of the AWA experimental setup: S, Q, and D indicate solenoids, quadrupoles, and dipoles,
respectively; the four integrating current transformers are labeled I1 through I4; the eight diagnostic ports are labeled
C1 through C8.
________________
*Work supported by the Department of Energy, Division of High Energy Physics, under contract W-31-109-ENG-38.
100
MeV electrons. The light is then reflected by a mirror and
leaves the vacuum chamber through a diagnostic view
port.
Figure 2 shows measurements of pulse length as a
function of bunch charge. In these plots the charge was
measured by the integrating current transformer I1 (see
Fig. 1), and the bunch length was measured using the
aerogel radiator and the streak camera. In Fig. 2a we
show a detailed charge scan in the range 15 - 35 nC and
the corresponding FWHM of the pulses; many pulses
were measured in this range and then binned in groups of
51. The plot shows the average value of charge and
bunch length for seven bins, indicating that in this charge
range the pulse length is almost constant, fluctuating
between 15 and 20 ps. The plot in Fig. 2b shows a charge
scan over a much wider range, but each point in this
graph is the average of only three pulses. We have plotted
the FWHM of the bunch lengths and also the 95% RMS
values (this was calculated taking into account only the
section of the pulse profile with intensity within 95% of
the peak value, with the purpose of discarding the effect
of the small background noise at the wings of the
distribution). The ratio between the 95% RMS and the
FWHM values shows that the pulses are not gaussian.
The large fluctuation in the FWHM of the pulses also
shows that the detailed shape of the temporal profile
varies considerably from pulse to pulse. In all of these
bunch length measurements the changes in bunch charge
are accomplished by varying the laser pulse energy
(either by purposely attenuating the laser beam, or due to
40
FWHM bunch length (ps)
pass through two standing-wave, π/2 mode linac tanks
(shunt impedance of 21.5 MΩ/m), increasing the beam
energy to about 14 MeV. The linac structures have large
irises (10.16 cm diameter) to minimize the generation of
wakefields by the propagation of the high charge drive
bunches [4]. The beam is then focused by quadrupoles
and bent by three dipoles to allow for the injection of
both the drive beam and the witness beam into the
wakefield experimental section. The witness gun [5] is a
6-cell standing-wave π/2 mode structure that generates 4
MeV bunches of 300 pC. Its photocathode material is
copper. The witness beam goes through combining
optics and then through the wakefield device (plasma or
dielectric structure) trailing the drive beam, thereby
probing the wakefields excited by the drive bunches.
Energy changes of the witness beam are measured by a
spectrometer magnet located downstream of the
wakefield experimental section.
The two RF guns and the two linac structures are
powered by a single klystron (Thomson TH2022D; 24
MW, 4 µs pulses), via necessary power splitters and
phase shifters. The laser system is comprised of a dye
oscillator (496 nm) pumped by a tripled YAG, which is
then followed by a dye amplifier, a doubling crystal and
finally an excimer amplifier. This laser is capable of
producing up to 8 mJ with 6 ps FWHM at 248 nm. The
laser beam is then split and a small fraction of it (~ 15%)
sent to the witness gun. There is an adjustable delay
between the drive gun and the witness gun laser beams.
This allows us to vary the delay between the drive and the
witness electron bunches (obviously the RF phases have
to be adjusted accordingly, in order to maintain the same
launching phase).
3 DIAGNOSTICS AND BEAM
CHARACTERIZATION
(a)
0
0
20
40
bunch charge (nC)
60
bunch length (ps)
There are four integrating current transformers (Bergoz
ICT - 082-070-20:1) installed on the beamlines to
measure bunch charge at various locations as indicated in
Fig. 1. Bunch length measurements [6] are performed
using radiation emitted by the electron beam as it passes
through a Cerenkov radiator. The Cerenkov light is sent
to a Hamamatsu M1952/C1587 streak camera for pulse
length measurements. Insertable quartz or aerogel plates
are used as Cerenkov radiators. The aerogels require a
more elaborate holder than the quartz plates, but offer
several advantages. The aerogels need to be in a vacuumtight holder which is inserted in the beam path. The
electron beam enters our holder through a thin aluminum
window (0.15 mm). It then traverses the aerogel (under
atmospheric pressure) and emits Cerenkov radiation,
which leaves the aerogel holder through a quartz window.
Our aerogels have an index of refraction of 1.009,
therefore the Cerenkov light is emitted at an angle of
7.4° with respect to the direction of propagation of the 14
20
FWHM
95% RMS
40
20
(b)
0
0
40
80
120
bunch charge (nC)
Figure 2: Measurements of electron bunch length as a
function of bunch charge: (a) detailed pulse length
measurement in the 15 - 35 nC range; (b) measurement
over a wider range of charges, each point represents the
average of three pulses.
101
4 WAKEFIELD MEASUREMENTS
We can map out the wakefields in a dielectric loaded
structure by varying the delay between the drive bunch
and the probe (witness) bunch. By measuring the change
in energy of the witness beam, we obtain a direct
measurement of the wake potential. This detailed
mapping of the wakefields [1] was performed in a
structure consisting of a hollow borosilicate glass
cylinder (ε ∼ 4) with inner radius of 5.0 mm and outer
radius of 7.7 mm. Figure 3 shows the wake potential as
measured from the energy change of the centroid of the
witness beam as a function of the delay between the two
bunches. The largest energy shift of the centroid is about
2.5 MeV/m (for drive bunches of 11 nC). Because the 2.5
mm FWHM length of the witness beam is a significant
fraction of the wavelength of the wakefield (in this
specific case the wakefield has an RF frequency of 15
GHz), the actual gradient is larger than the gradient
measured from the witness beam centroid change in
energy. By comparing the data with numerical
simulations convolved with a Gaussian witness bunch
shape, a true gradient of 3.6 MeV/m is inferred.
Additional optimization of the laser injection phase and
beamline magnet settings resulted in a gradient of
approximately 6 MeV/m with a 20 nC drive beam
intensity.
transformer configuration. The drive beam generates an
RF pulse as it passes through one of the structures. This
RF pulse is coupled to the second dielectric structure (via
a waveguide) where the witness beam is accelerated. The
second tube has the same fundamental frequency but
lower group velocity and transverse dimensions, thus
providing an accelerating field step-up by compressing
the RF pulse. Multiple drive bunches can be used, spaced
by an integral multiple of the RF period, to provide a long
accelerating pulse. Another advantage of having two
separate dielectric structures is that the drive structure can
be designed with sufficiently low transverse impedance to
avoid beam breakup problems.
We have recently been designing and building
dielectric wakefield transformers [7] operating at 7.8 GHz
and also at 15.6 GHz. We have also succeeded in
generating multiple drive bunches, by optically splitting
and appropriately delaying the laser pulse to the drive
gun. Figure 4 shows the envelope of the RF macropulse
generated by a bunch train consisting of 4 bunches
separated by 3.077 µs.
2.5
2.0
RF power (MW)
the natural fluctuations in the laser power); when the
change in the emitted bunch charge is large, it becomes
necessary to readjust the solenoids and RF phases to
compensate for the changes in the space charge forces.
1.5
1.0
0.5
0.0
0
10
20
30
40
50
time (ns)
Figure 4: RF macropulse envelope for train of four drive
bunches in a 7.8 GHz structure.
6 REFERENCES
Figure 3: Wake potential measurement for a 15 GHz
dielectric structure. Each data point is the change in the
bend view centroid of the witness beam at the
spectrometer 60° port.
[1] P. Schoessow et al., J. Appl. Phys. 84, 663 (1998).
[2] N. Barov et al., Phys. Rev. Lett. 80, 81 (1998).
[3] W. Gai et al., Proceedings of PAC89, Chicago, IL,
612 (1989).
[4] E. Chojnacki et al., Proceedings of PAC93,
Washington, D.C., 815 (1993).
[5] J.G. Power et al., Rev. Sci. Instrum. 69, 1295 (1998).
[6] M.E. Conde et al., submitted to Phys. Rev. Special
Topics - Accelerators and Beams.
[7] M.E. Conde et al., submitted to Proceedings of
Advanced Accelerator Concepts, Baltimore, MD,
(1998).
5 DIELECTRIC WAKEFIELD
TRANSFORMER
There are advantages in building a dielectric wakefield
accelerator with two separate dielectric structures, in a
102